Back at the beginning of the year, in the before times, I wrote that APIs are for life, and that if you change something, even if you make it better, you're likely to have unhappy customers. And that's true even if what you've changed isn't a public API.
One of the things we did with Combat Flight Simulator 3 (CFS3) was improve the graphics. We took advantage of newer hardware to increase resolution and fidelity. We did this in two main areas. The aircraft themselves and the terrain engine. We knew that the FlightSim community had already created hundreds of aircraft, and we wanted to allow them to be used, so while we improved rendering and animations on the aircraft we made sure to handle last years formats as well. Since CFS3 had a lot more air to ground action than earlier versions we also wanted to make sure that we could get sufficient detail to get a sense of speed and action everywhere. Because we started with the FlightSim codebase and data set we already had the entire world covered, so we didn't worry about add-on scenery.
According to our customers, we chose poorly. Even though we didn't officially support 3rd party updates to the scenery, a small portion of the customer base had figured out how to add scenery in earlier versions. They made changes not only for themselves, they shared those mods with others. There were even a few companies that had commercial products with that scenery. Which meant that when those add ons didn't work we didn't just have a few hard-core fans who were upset. All of their customers were upset with us too.
This was in the days of box products, so it was too late to do anything for CFS3, and that bad taste certainly impacted sales. Because of the low sales and reduced profit (CFS still made money, but not as much as expected) CFS4 was cancelled. Was that entirely because we changed a private API? Of course not. The entire market was down, flight sims were losing favor, and Microsoft was focusing on higher return on investment. But if we hadn't made that change and had happier customers and better press things might have turned out differently.
We did take the lesson to heart though, and while we incorporated many of the improvements to the terrain engine into the next version of Flight Simulator we took the extra time to ensure that the improvements worked with the old formats instead of replacing them. Flight Simulator was much better received, and managed to survive for 2 more versions and some add-ons over 5 years before it too was shut down for not having a high enough profit.
So remember, surprising your customers isn't always a good thing.